WO2017157119A1

WO2017157119A1 - Method and device for identifying abnormal behavior of vehicle

Info

Publication number: WO2017157119A1
Application number: PCT/CN2017/073477
Authority: WO
Inventors: 马昌军; 解海波
Original assignee: 中兴通讯股份有限公司
Priority date: 2016-03-18
Filing date: 2017-02-14
Publication date: 2017-09-21
Also published as: CN107204114A

Abstract

A method and device for identifying an abnormal behavior of a vehicle. The identification method comprises: obtaining vehicle monitoring data collected on multiple monitoring points arranged in a statistics region, and extracting behavior characteristic data of a to-be-inspected vehicle from the vehicle monitoring data (101); obtaining a traveling track sequence of the to-be-inspected vehicle according to the behavior characteristic data (102); and determining whether the to-be-inspected vehicle has an abnormal behavior according to the traveling track sequence and a pre-obtained vehicle behavior mode in the statistics region (103). The identification device correspondingly comprises a first obtaining module (401), a second obtaining module (402), and a determining module (403). The method and the device resolve the defect in the related art of failure to perform correlation analysis on vehicle monitoring data collected on multiple monitoring points, and also resolve the problem of failure to automatically determine abnormal behaviors of a vehicle in a long time, thereby improving the identification accuracy and the identification efficiency of the abnormal behaviors of the vehicle.

Description

Method and device for identifying abnormal behavior of vehicles

Technical field

The invention relates to the field of intelligent traffic monitoring, in particular to a method and a device for identifying abnormal behavior of a vehicle.

Background technique

The use of traffic information collection technology for traffic violation detection and real-time monitoring of traffic conditions has been widely used. The existing technology can realize automatic detection and photo-taking for illegal activities such as red light, speeding, illegal parking and reverse driving. In addition, for the abnormal behavior of vehicles in a long time or space span, such as: occlusion number plate, conversion number plate and deck, etc., it is also necessary to perform correlation analysis on time-discontinuous data taken by multiple cameras. The current common solutions for such needs are:

First, manual retrieval and manual filtering: for example, deck vehicle identification, usually the real owner of the number plate finds that the vehicle is abnormally violated and reports to the traffic management department, or the number of violations of the deck vehicle is too much, causing the traffic control department to pay attention, manually retrieved All the monitoring records of the number plate, and manually check the illegal vehicles. However, when manually performing manual retrieval and filtering, the efficiency is very limited, and it is impossible to correlate the massive monitoring data of multiple cameras.

Secondly, by analyzing the time and space relationship of the vehicle, that is, the vehicle passes the minimum transit time between the bayonet to determine whether the deck is: if the time difference of a vehicle being photographed at the two bayonet is less than a certain time threshold (the minimum between the corresponding bayonet) Through time), one of the cars is considered to be a deck car. However, the method of determining the deck through the minimum transit time between the bayonet is difficult to accurately set the minimum passing time, and there is a large probability of misjudgment. In addition, if the deck vehicle is different from the real vehicle, it cannot be detected. A certain probability of missed judgment.

Although the above two schemes can find some abnormal behaviors of the vehicle, the above two schemes have the following disadvantages. Among them, manual manual retrieval and filtering are very inefficient, and it is impossible to correlate the massive monitoring data of multiple cameras. . In addition, the way to determine the deck through the minimum transit time between the bayonet is difficult to accurately set the minimum passing time, and there is a large misjudgment. Probability, in addition, if the deck vehicle does not appear when it is different from the real vehicle, there is a certain probability of missed judgment. In addition, if the vehicle blocks or replaces the license plate for a certain period of time, the number plate is exchanged for the real license plate before passing through the toll booth and the traffic inspection intersection. Unless the card changing process is taken, such behavior is difficult to find or Judge.

In summary, the prior art has the problem that it is impossible to correlate the vehicle monitoring data captured by a plurality of cameras, that is, it is impossible to identify and judge the abnormal behavior of the vehicle over a long period of time.

Summary of the invention

In order to solve the problem that the abnormal behavior of the vehicle cannot be recognized and judged in a long time in the prior art, the embodiment of the invention provides a method and a device for identifying abnormal behavior of the vehicle.

In order to solve the above technical problem, an embodiment of the present invention provides a method for identifying an abnormal behavior of a vehicle, where the identification method includes:

Obtaining vehicle monitoring data collected by a plurality of monitoring points set in the statistical area, and extracting behavior characteristic data of the vehicle to be inspected from the vehicle monitoring data;

Obtaining a travel trajectory sequence of the vehicle to be inspected according to the behavior characteristic data;

Determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the vehicle behavior pattern in the statistical region acquired in advance.

Optionally, the behavior characteristic data includes a license plate number and a monitoring point identifier, and the acquiring the traveling trajectory sequence of the to-be-tested vehicle according to the behavior characteristic data includes: according to the license plate number of the vehicle to be inspected The vehicle to be tested passes the time sequence of the plurality of monitoring points in the statistical area, and records the monitoring point identifier corresponding to the plurality of monitoring points that the vehicle to be inspected sequentially passes as the traveling track sequence of the vehicle to be inspected.

Optionally, before the determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence of the vehicle to be inspected and the vehicle behavior pattern in the statistical region acquired in advance, the identifying method further includes: Pre-acquiring vehicle monitoring data of all vehicles in the statistical area to obtain a vehicle behavior pattern, wherein the vehicle behavior pattern includes: a reachability relationship between monitoring points, a transit time between monitoring points, and a vehicle Driving track class template.

Optionally, the reachability relationship between the monitoring points is determined according to the following manner: acquiring, according to vehicle monitoring data of all vehicles in the statistical area, a traveling track sequence of all the vehicles Columns; performing statistical analysis on the trajectory sequences of all the vehicles to obtain a normal trajectory sequence, and determining adjacent monitoring point identifiers in the normal trajectory sequence; according to the adjacent monitoring point identifiers, obtaining direct The monitoring point of the relationship, wherein the previous monitoring point indicated by the adjacent monitoring point identifier can directly reach the next monitoring point in the adjacent monitoring point identifier.

Optionally, the transit time between the monitoring points is determined according to the following manner: calculating, according to the vehicle monitoring data, an average transit time between monitoring points having a direct reachability relationship within a preset time period, and The average transit time is recorded as the transit time between the monitoring points within the preset time period.

Optionally, after the acquiring the traveling trajectory sequence of the to-be-tested vehicle according to the behavior characteristic data, the identifying method further includes: modifying the license plate number error data of the to-be-tested vehicle, specifically: according to: Determining a reachable relationship between the monitoring points, determining whether there is a direct reachability relationship between the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle to be inspected; if not, acquiring the All the paths between the monitoring points in the statistical area that do not have a direct reachability relationship, wherein the path is represented by a sequence consisting of monitoring point identifiers corresponding to the monitoring points in sequence; according to the monitoring points in all the paths a bit identifier, the monitoring point identifier of the abnormal monitoring data monitored in the all paths, wherein the abnormal monitoring data includes at least: the monitored license plate number does not match the vehicle registration information, the license plate number registration information, and the path phase The monitoring point corresponding to the neighbor monitoring point identifier does not have a direct reachability relationship and the monitoring point with direct reachability in the path does not The transit time between the monitoring points; the monitoring point identifier of the monitored abnormal monitoring data is merged into the traveling track sequence of the vehicle to be inspected, and it is determined that the adjacent monitoring point identifiers in the merged traveling track sequence correspond to Whether the monitoring point meets the transit time between the monitoring points; if it is satisfied, and the similarity between the monitored license plate number and the license plate number of the vehicle to be inspected is greater than the first preset value, the monitored license plate will be The number is corrected.

Optionally, determining, according to the traveling trajectory sequence of the vehicle to be inspected and the vehicle behavior pattern in the statistical region that is acquired in advance, whether the vehicle to be inspected has an abnormal behavior, including: according to the traveling trajectory sequence, The reachability relationship between the monitoring points and the communication between the monitoring points Determining whether the vehicle to be inspected has an abnormal behavior; and/or determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the driving trajectory class template.

Optionally, the determining, according to the traveling trajectory sequence, the reachable relationship between the monitoring points, and the transit time between the monitoring points, determining whether the vehicle to be inspected has an abnormal behavior, including: according to the driving track The sequence, the reachability relationship between the monitoring points, and the transit time between the monitoring points, and sequentially determine whether the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle to be inspected have direct reachability relationship And whether the transit time between the monitoring points is met; if the monitoring point corresponding to the adjacent monitoring point identifier does not have a direct reachable relationship, or the monitoring point corresponding to the adjacent monitoring point identifier has a direct reachable relationship, If the transit time between the monitoring points is satisfied, it is determined that the vehicle to be tested has an abnormal behavior.

Optionally, determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the driving trajectory template comprises: splitting the traveling trajectory sequence of the to-be-tested vehicle according to a preset time interval a plurality of driving track sub-sequences; acquiring a driving track class template having the first monitoring point identifier and the last monitoring point identifier according to the first monitoring point identifier and the last monitoring point identifier of the driving track sub-sequence Calculating a similarity between the travel track subsequence and the acquired travel track type template, and if the similarity is less than the second preset value, determining that the vehicle to be tested has an abnormal behavior.

According to another aspect of the embodiments of the present invention, an embodiment of the present invention further provides an apparatus for identifying an abnormal behavior of a vehicle, where the identifying apparatus includes:

a first acquiring module, configured to acquire vehicle monitoring data collected by a plurality of monitoring points set in the statistical area, and extract behavior characteristic data of the to-be-tested vehicle from the vehicle monitoring data;

a second acquiring module, configured to acquire a traveling trajectory sequence of the to-be-tested vehicle according to the behavior characteristic data;

The determining module is configured to determine whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the vehicle behavior pattern in the statistical region acquired in advance.

Optionally, the behavior characteristic data includes a license plate number and a monitoring point identifier, and the second acquiring module is further configured to: according to the license plate number of the vehicle to be inspected and the plurality of monitoring in the statistical area of the vehicle to be inspected The chronological order of the points, the multiple monitoring points that the vehicle to be tested passes in sequence The corresponding monitoring point identification is recorded as the traveling trajectory sequence of the vehicle to be inspected.

Optionally, the identifying apparatus further includes a third acquiring module, configured to acquire a vehicle behavior mode according to pre-acquired vehicle monitoring data of all vehicles in the statistical area, where the vehicle behavior mode includes: monitoring a position The reachable relationship between the two, the transit time between the monitoring points, and the trajectory template of the vehicle.

Optionally, the third acquiring module is further configured to: acquire, according to vehicle monitoring data of all vehicles in the statistical area, a traveling trajectory sequence of all the vehicles; perform statistical analysis on the traveling trajectory sequence of all the vehicles, Obtaining a normal driving trajectory sequence, and determining an adjacent monitoring point identifier in the normal driving trajectory sequence; and obtaining, according to the adjacent monitoring point identifier, a monitoring point having a direct reachability relationship, wherein the adjacent monitoring point The previous monitoring point indicated by the bit identifier can directly reach the next monitoring point in the adjacent monitoring point identifier.

Optionally, the third obtaining module is further configured to calculate, according to the vehicle monitoring data, an average transit time between monitoring points having a direct reachability relationship within a preset time period, and the average The transit time is recorded as the transit time between the monitoring points within the preset time period.

Optionally, the identifying device further includes a correction module configured to correct the license plate number error data of the to-be-tested vehicle, and further configured to: determine, according to the reachable relationship between the monitoring points Obtaining a direct reachability relationship between the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle; if not, acquiring between the monitoring points in the statistical area that do not have a direct reachable relationship All the paths, wherein the path is represented by a sequence consisting of monitoring point identifiers corresponding to the monitoring points in sequence; and monitoring the monitored abnormal monitoring data in all the paths according to the monitoring point identifiers in all the paths a point identifier, wherein the abnormality monitoring data includes at least: the monitored license plate number does not match the vehicle registration information, the license plate number registration information, and the monitoring point corresponding to the adjacent monitoring point identifier in the path does not have a direct reachable relationship And the monitoring point with direct reachability in the path does not meet the transit time between the monitoring points; the abnormal monitoring data will be monitored The monitoring point identifier is merged into the traveling track sequence of the vehicle to be inspected, and it is determined whether the monitoring point corresponding to the adjacent monitoring point identifier in the merged traveling track sequence satisfies the transit time between the monitoring points; And monitoring the similarity between the license plate number and the license plate number of the vehicle to be inspected is greater than the first preset value, and then monitoring The license plate number to be corrected.

Optionally, the determining module is further configured to: determine, according to the traveling trajectory sequence, the reachability relationship between the monitoring points, and the transit time between the monitoring points, whether the vehicle to be tested has an abnormal behavior; And determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the driving trajectory class template.

Optionally, the determining module includes a first determining unit, configured to sequentially determine, according to the traveling trajectory sequence, a reachable relationship between monitoring points, and a transit time between monitoring points, sequentially determining the vehicle to be inspected Whether the monitoring point corresponding to the adjacent monitoring point identifier in the traveling track sequence has a direct reachability relationship and whether the transit time between the monitoring points is met; if the monitoring point corresponding to the adjacent monitoring point identifier does not have direct reachability The relationship, or the monitoring point corresponding to the adjacent monitoring point identifier has a direct reachable relationship but does not satisfy the transit time between the monitoring points, and then determines that the vehicle to be tested has an abnormal behavior.

Optionally, the determining module further includes a second determining unit, configured to split the traveling trajectory sequence of the to-be-tested vehicle into a plurality of driving trajectory sub-sequences according to a preset time interval; a first monitoring point identifier and a last monitoring point identifier of the sequence, obtaining a driving trajectory class template having the first monitoring point identifier and a last monitoring point identifier; calculating the driving trajectory subsequence and the obtained driving The similarity between the trajectory class templates, if the similarity is less than the second preset value, determining that the vehicle to be tested has an abnormal behavior.

The beneficial effects of the embodiments of the present invention are:

A method for identifying an abnormal behavior of a vehicle according to an embodiment of the present invention first acquires vehicle monitoring data collected by a plurality of monitoring points set in a statistical area, and extracts behavior characteristic data of the vehicle to be inspected from the vehicle monitoring data, and then according to the vehicle The behavior characteristic data acquires a traveling trajectory sequence of the vehicle to be inspected, and finally determines whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the pre-acquired vehicle behavior pattern. The embodiment of the invention solves the defect that the vehicle monitoring data collected by the plurality of monitoring points cannot be correlated in the prior art by analyzing the traveling trajectory sequence of the vehicle to be inspected, and solves the problem that the vehicle cannot be used for a long time. The problem of automatic judgment of abnormal behavior increases the recognition accuracy and recognition efficiency of abnormal behavior of vehicles.

DRAWINGS

1 is a flow chart showing the steps of a method for identifying an abnormal behavior of a vehicle in a first embodiment of the present invention;

2 is a flow chart showing the steps of obtaining a reachability relationship between monitoring points in the second embodiment of the present invention;

Figure 3 is a schematic view showing the arrangement of monitoring points of a traffic intersection;

Figure 4 is a diagram showing the reachability relationship between the respective monitoring points in Figure 3;

Figure 5 is a flow chart showing the steps of correcting the license plate number error data of the vehicle to be inspected in the third embodiment of the present invention;

Fig. 6 is a block diagram showing the configuration of an apparatus for identifying an abnormal behavior of a vehicle in a fifth embodiment of the present invention.

detailed description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While the embodiments of the present invention have been shown in the drawings, the embodiments Rather, these embodiments are provided so that this disclosure will be more fully understood and the scope of the disclosure will be fully disclosed.

First embodiment:

FIG. 1 is a flow chart showing the steps of a method for identifying an abnormal behavior of a vehicle according to a first embodiment of the present invention, the method includes:

Step 101: Acquire vehicle monitoring data collected by a plurality of monitoring points set in the statistical area, and extract behavior characteristic data of the vehicle to be inspected from the vehicle monitoring data.

In this step, a plurality of monitoring points set in the statistical area can capture an image of the vehicle, process and optically recognize the image of the vehicle, and collect vehicle monitoring data of the vehicle. Specifically, the monitoring point continuously collects the image of the vehicle passing the monitoring point, and extracts the vehicle license plate number, the vehicle model, the vehicle color and other characteristic information, and establishes a monitoring database in the statistical area. Optionally, the vehicle monitoring data in the monitoring database may include not only dynamic data such as the license plate number of the vehicle passing through the monitoring point and the passing time of the monitored point, but also the monitoring point identification of the monitoring point and the monitoring point. Static data such as location. In addition, in this step, you can also The behavior characteristic data of the vehicle to be inspected is extracted from the vehicle monitoring data, wherein the behavior characteristic data includes at least: a license plate number and a monitoring point identifier corresponding to the monitoring point.

Step 102: Acquire a travel trajectory sequence of the vehicle to be inspected according to the behavior characteristic data.

Specifically, the behavior characteristic data includes a license plate number and a monitoring point identifier. In this step, the vehicle to be inspected may be sequentially according to the license plate number of the vehicle to be inspected and the time sequence of the plurality of monitoring points in the statistical area of the vehicle to be inspected. The monitoring point identification corresponding to the plurality of monitored points is recorded as the traveling track sequence of the vehicle to be inspected.

Step 103: Determine whether an abnormal behavior exists in the vehicle to be inspected according to the driving trajectory sequence and the vehicle behavior pattern in the statistical area acquired in advance.

In this step, the vehicle behavior pattern in the statistical area may be acquired in advance, and then the vehicle to be inspected is determined to have abnormal behavior according to the traveling trajectory sequence of the vehicle to be inspected and the vehicle behavior pattern in the statistical area acquired in advance. Specifically, the vehicle behavior pattern in the statistical area may include a reachability relationship between the monitoring points in the statistical area, a transit time between the monitoring points, and a driving trajectory template of the vehicle. Abnormal behavior of the vehicle may include behaviors such as vehicle decks, vehicle flops, obstruction of vehicle number plates, and illegal driving.

In this embodiment, the traveling trajectory sequence of the vehicle to be inspected is analyzed, and according to the driving trajectory sequence and the vehicle behavior pattern in the pre-acquired statistical region, whether the vehicle in question is abnormal or not is determined, and the multiple solutions in the prior art cannot be solved. The vehicle monitoring data collected by the monitoring point is used for correlation analysis, and solves the problem that the vehicle abnormal behavior cannot be automatically judged for a long time, and the recognition accuracy and recognition efficiency of the abnormal behavior of the vehicle to be tested are increased.

Second embodiment:

Before determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the vehicle behavior pattern in the pre-acquired statistical region, it is also necessary to acquire the vehicle behavior pattern in the statistical region. Specifically, the vehicle behavior mode may be obtained according to vehicle monitoring data of all vehicles in the pre-acquired statistical area, wherein the vehicle behavior mode includes: a reachability relationship between the monitoring points, a transit time between the monitoring points, and a vehicle Driving track class template. In addition, when the vehicle monitoring data is collected in real time, the vehicle monitoring data acquired in real time can also be used as a basis for acquiring the behavior pattern of the vehicle.

The vehicle behavior pattern is explained below.

Specifically, the vehicle is photographed by a plurality of monitoring points along the road during driving to form vehicle monitoring data. Through the analysis and research on the massive historical vehicle monitoring data, it can be found that there is a certain time and space regularity of the vehicle's motion, and these laws form a certain vehicle behavior pattern. Specifically, each monitoring point can be regarded as a state, and the monitoring point identifier corresponding to all monitoring points in the statistical area constitutes a finite state space, and the finite state space can be expressed as E={1, 2, ...,N}. When the vehicle is photographed by different monitoring points, it can be considered that the vehicle is transferred between different states, wherein the vehicle running process can be described by using the monitoring point identifier corresponding to the monitoring point passed by the vehicle, for example, for example, for example, The sequence of the trajectory of the A vehicle in one driving process can be represented by A=(A1, A2, ..., An). Then according to the statistical principle, the transition probability of the vehicle between different states can be obtained. In this way, according to the vehicle monitoring data of all the vehicles in the pre-acquired statistical area, the vehicle behavior patterns such as the reachability relationship between the monitoring points, the transit time between the monitoring points, and the traveling trajectory template of the vehicle can be obtained. Specifically, when acquiring the vehicle behavior mode, the vehicle monitoring data of some vehicles in the statistical area may also be sampled, but in order to avoid sampling inaccuracy, it is preferable to analyze the vehicle monitoring data of all vehicles in the statistical area to obtain the vehicle. Behavior pattern.

The following describes the method for determining the reachability relationship between the monitoring points in the vehicle behavior mode, the transit time between the monitoring points, and the driving trajectory class template of the vehicle.

As shown in FIG. 2, a flow chart of steps for determining a reachability relationship between monitoring points in a second embodiment of the present invention includes:

Step 201: Acquire a travel trajectory sequence of all vehicles according to vehicle monitoring data of all vehicles in the statistical area.

In this step, specifically, according to the vehicle monitoring data of all vehicles in the statistical area, the behavior characteristic data of all the vehicles can be acquired, and then the traveling trajectory sequence of all the vehicles is obtained according to the behavior characteristic data of all the vehicles, that is, according to the license plate of the vehicle. The number and the monitoring point identifier corresponding to the monitoring point that the vehicle passes in sequence acquire the traveling track sequence of the vehicle.

Step 202: Perform statistical analysis on the traveling trajectory sequences of all the vehicles to obtain a normal driving trajectory sequence, and determine adjacent monitoring point identifiers in the normal driving trajectory sequence.

In this step, specifically, when performing statistical analysis on the traveling trajectory sequence of all the vehicles, if the number of occurrences of a traveling trajectory sequence is less than a preset threshold, the traveling trajectory sequence is regarded as an abnormal traveling trajectory sequence; otherwise, The travel trajectory sequence is considered to be a normal travel trajectory sequence. Specifically, the preset threshold may be determined according to the total number of driving trajectories and the total type of the traveling trajectory sequence. For example, the average number of occurrences of a travel trajectory sequence may be obtained according to the ratio of the total number of travel trajectory sequences to the total trajectory sequence, and then the preset threshold may be set according to the average number of occurrences, for example, according to the average number of occurrences and one The product of the scaling factor determines the preset threshold. Optionally, the scaling factor may be a value less than 1, such as 0.1, 0.01, and the like.

Step 203: Obtain a monitoring point with a direct reachability relationship according to the adjacent monitoring point identifier.

In this step, the monitoring point with the direct reachability relationship can be obtained according to the adjacent monitoring point identifier. Specifically, the previous monitoring point indicated by the adjacent monitoring point identifier can directly reach the adjacent monitoring point identifier. The next monitoring point in the middle.

In addition, the monitoring point having the direct reachability relationship is explained as follows: if the first monitoring point can directly reach the second monitoring point without passing through the third monitoring point, the first monitoring point and the second monitoring can be determined. The point has a direct reachability relationship; if the first monitoring point must pass the third monitoring point to reach the second monitoring point, it is determined that the first monitoring point and the second monitoring point do not have a direct reachable relationship. The first monitoring point, the second monitoring point, and the third monitoring point are different from each other, and are any monitoring points of the plurality of monitoring points in the statistical area.

The following provides an example of the reachability relationship between monitoring points.

As shown in FIG. 3, it is a schematic diagram of the arrangement of monitoring points of a traffic intersection. In FIG. 3, A, B, C, D, E, F, G, and H are the monitoring point identifiers of the monitoring points, and it can be seen from FIG. 3 that the road 1 prohibits the U-turn. Assuming that the normal traveling trajectory sequence is FGHC, it can be concluded that the monitoring points corresponding to the adjacent monitoring point identifiers F and G, G and H, H and C are monitoring points having a direct reachable relationship. By analogy, the adjacent monitoring point identifiers in all normal driving trajectory sequences can be obtained, and the reachability relationship between the respective monitoring points is obtained. Specifically, as shown in FIG. 4, it is a reachability relationship diagram between each monitoring point in FIG. 3, where the orientation between the two monitoring point identifiers is The connection indicates that there is a direct reachability relationship between the monitoring points corresponding to the two monitoring point identifiers. In addition, the connection direction also indicates that the transfer path between the two monitoring points is directed.

Specifically, after obtaining the reachability relationship between the monitoring points, the matrix P=(p _ij ) _n×n can also be used to record the reachability relationship between the monitoring points. P denotes the reachability relationship matrix between the monitoring points, n denotes the monitoring point identifier corresponding to all monitoring points in the statistical area, and p _ij denotes the monitoring point corresponding to the monitoring point identifier of the i-th row in the P matrix The j-th column monitoring point identifies the reachability relationship between the corresponding monitoring points. Wherein, when p _ij =1, it indicates that the monitoring point corresponding to the monitoring point identifier of the i-th row in the P matrix has a direct reachable relationship with the monitoring point corresponding to the monitoring point identifier of the j-th column; when p _ij =0 , indicating that the monitoring point corresponding to the monitoring point identifier of the i-th row in the P matrix does not have a direct reachable relationship with the monitoring point corresponding to the monitoring point identifier of the j-th column. In addition, it should be noted that, in the matrix, the reachability relationship between the two monitoring point identifiers corresponding to the monitoring points is directed, for example, in p _ij , the i-th row monitoring point identifier corresponds to The monitoring point is the previous monitoring point indicated by the adjacent monitoring point identifier, and the monitoring point corresponding to the monitoring point identifier of the jth column is the next monitoring point indicated by the adjacent monitoring point identifier.

In addition, in this embodiment, optionally, when determining the transit time between the monitoring points, the average between the monitoring points having a direct reachable relationship within a preset time period may be calculated according to the vehicle monitoring data. The transit time, and the average transit time is recorded as the transit time between the monitoring points within the preset time period.

Specifically, since the time between the monitoring points of the vehicle passing through the direct reachability relationship varies with the traffic condition, that is, the transit time between the monitoring points has a tidal characteristic, and thus can be separately calculated according to different preset time periods. Average transit time between monitoring points with direct reachability. For example, if 70 minutes is used as a preset time period and a certain amount of overlap can be reserved between different preset time segments, 0:00~1:10 can be divided into a preset time period, 1:00~ 2:10 is a preset time period, and so on, can get different preset time periods.

Of course, after obtaining the transit time between the monitoring points, the matrix can also be used to record the transit time between the monitoring points within a preset time period. For example, the matrix T=(t _ij ) _n×n can be used to record the transit time between the monitoring points within a preset time period. Wherein, T represents a transit time matrix between monitoring points in a preset time period, n represents a monitoring point identifier corresponding to all monitoring points in the statistical area, and t _ij represents an i-th row monitoring point identifier in the T matrix. The transit time between the corresponding monitoring point and the monitoring point corresponding to the monitoring point identifier of the jth column. Wherein, when the monitoring point corresponding to the monitoring point identifier of the i-th row in the T matrix has a direct reachable relationship with the monitoring point corresponding to the monitoring point identifier of the j-th column, the value of t _ij is the transit time; when in the T matrix When the monitoring point corresponding to the i-th row monitoring point identifier does not have a direct reachability relationship with the monitoring point corresponding to the j-th column monitoring point identifier, the value of t _ij is infinite.

In addition, in the embodiment, optionally, when acquiring the trajectory class template of the vehicle in the vehicle behavior mode, a clustering algorithm may be used to perform cluster analysis on all the normal trajectory sequences to obtain a trajectory of the vehicle. template.

Specifically, in the process of clustering all the normal trajectory sequences by using the clustering algorithm to obtain the trajectory class template of the vehicle, the Map function and the Reduce function can be used in the map-reduce method. The trajectory sequence of the vehicle: in the Map function, each vehicle monitoring data of each monitoring point is obtained, and the license plate number of a vehicle is taken as the key value, the time of the vehicle passing through each monitoring point, the monitoring point identification, Other information such as vehicle model and vehicle color is passed as a value to the Reduce function; in the Reduce function, the data of the same key value can be sorted according to the elapsed time, thereby obtaining the traveling trajectory sequence of the vehicle. In addition, when obtaining the travel trajectory type template of the vehicle, the distance between the travel trajectories corresponding to all the travel trajectory sequences can be measured by the Hausdorff distance, and then the trajectory similarity between the travel trajectories can be obtained. It is assumed that if there are m traveling trajectory sequences, a similarity matrix S=(S _ij ) _m×m can be obtained, where S _ij represents the corresponding trajectory of the i-th row in the similarity matrix S and the corresponding j-th column. The trajectory similarity between the travel trajectories. Finally, after obtaining the trajectory similarity of all the driving trajectory sequences, the spectral trajectory algorithm can be used to divide the driving trajectory corresponding to the m driving trajectory sequences into q driving trajectory class templates. Preferably, the clustering algorithm can adopt the K-means ( K-means) clustering algorithm.

In this embodiment, the vehicle behavior mode is trained according to the vehicle monitoring data of all the vehicles in the statistical area, and the accuracy of the vehicle behavior mode is increased, so that whether the vehicle to be inspected is determined according to the traveling trajectory sequence of the vehicle to be inspected and the vehicle behavior pattern When there is abnormal behavior, the identification of the abnormal behavior of the vehicle to be tested is more accurate.

Third embodiment:

In the vehicle monitoring data collected at the acquired monitoring points, due to various factors such as low image quality, image recognition error, or network transmission, the vehicle monitoring data may be partially wrong or incomplete, for example, in the license plate number. "D" and "0", "L", "T" and "1" and other relatively close numbers have a certain probability to recognize the wrong result. Such partial or incomplete vehicle monitoring data can affect the acquisition of vehicle behavior patterns and the identification of abnormal vehicle behavior. Therefore, after acquiring the traveling trajectory sequence of the vehicle to be inspected according to the behavior characteristic data, it is also necessary to correct the license plate number erroneous data of the vehicle to be inspected. Specifically, if there is no direct reachable relationship between the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle to be inspected, but the transit time between the monitoring points can be met, the vehicle may be When the monitoring point of the passing point is incorrectly recognized by the license plate number, the vehicle monitoring data is lost or the error information is combined. At this time, the abnormality monitoring data can be screened and filtered to correct the incorrect data of the license plate number.

Specifically, as shown in FIG. 5, a flow chart of steps for correcting the license plate number error data of the vehicle to be inspected according to the third embodiment of the present invention includes:

Step 301: Determine, according to the reachability relationship between the monitoring points, whether there is a direct reachability relationship between the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle to be inspected.

In this step, assuming that the traveling trajectory sequence of the vehicle to be inspected is A=(A1, A2, ..., An), the adjacent monitoring points in the traveling trajectory sequence A can be judged according to the reachable relationship between the monitoring points. It is determined whether there is a direct reachability relationship between the corresponding monitoring points. If there is no direct reachable relationship, the process proceeds to step 302.

Step 302: If not, obtain all paths between the monitoring points in the statistical area that do not have a direct reachability relationship.

In this step, specifically, if there is no direct reachable relationship between the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle to be inspected, the monitoring in the statistical area does not have a direct reachable relationship. All paths between points, wherein the path can be represented by a sequence consisting of monitoring point identifiers corresponding to the monitored points in sequence. For example, if there is no direct reachable relationship between the monitoring points corresponding to the adjacent monitoring point identifiers A1 and A2 in the traveling track sequence A, then The adjacent monitoring points identify all paths between the monitoring points corresponding to A1 and A2. Suppose that there are m total paths, and the total number of paths from A1 to A2 in the i-th path needs to pass through z monitoring points. Then the path P={P1, P2, ..., Pm}, path Pi=(A1, Ai1, ... , Aiz, A2), where m represents the number of all paths between the monitoring points corresponding to the adjacent monitoring point identifiers A1 and A2, and Pi represents the monitoring point between the adjacent monitoring point identifiers A1 and A2. The i-th path, Aiz represents the monitoring point identifier in the i-th path between the monitoring points corresponding to the adjacent monitoring point identifiers A1 and A2.

Step 303: Acquire, according to the monitoring point identifiers in all the paths, the monitoring point identifiers of the abnormal monitoring data monitored in all the paths.

In this step, specifically, the abnormal monitoring data includes at least the monitored license plate number does not match the vehicle registration information, the license plate number registration information, and the monitoring point corresponding to the adjacent monitoring point identifier in the path does not have a direct reachable relationship. The monitoring point with direct reachability in the path does not meet the transit time between the monitoring points, that is, when there is monitoring point monitoring, the license plate number does not match the vehicle registration information, or there is no monitored license plate number registration information, or path The monitoring points corresponding to the adjacent monitoring point identifiers do not have a direct reachable relationship, or the monitoring points corresponding to the adjacent monitoring point identifiers in the path have a direct reachable relationship but do not satisfy the monitoring point. When the transit time is obtained, the monitoring point identifiers corresponding to the monitoring points are obtained.

Step 304: The monitoring point identifier of the monitored abnormality monitoring data is merged into the traveling track sequence of the vehicle to be inspected, and it is determined whether the monitoring point corresponding to the adjacent monitoring point identifier in the merged traveling track sequence satisfies the monitoring point. The transit time between.

In this step, specifically, if the monitoring point corresponding to the adjacent monitoring point identifier in the merged traveling track sequence satisfies the transit time between the monitoring points, the process proceeds to step 305.

Step 305: If the monitoring point corresponding to the adjacent monitoring point identifier in the merged traveling track sequence satisfies the transit time between the monitoring points, and the similarity between the monitored license plate number and the license plate number of the vehicle to be inspected is greater than The first preset value corrects the monitored license plate number.

In this step, if the monitoring point between the monitoring points corresponding to the adjacent monitoring point identifiers in the merged traveling track sequence meets the transit time between the monitoring points, and the monitored license plate number and the license plate number of the vehicle to be inspected If the similarity is greater than the first preset value, the monitored license plate number is corrected. Optionally, after the monitored license plate number is corrected, a manual confirmation can be submitted. In addition, for a license plate number character that easily causes erroneous data, a matching character having a similar shape can be manually set, thereby obtaining a correct character based on the identification data.

For example, the "Su A23F45" and the "Su A23P45" are the fifth character in the two license plate numbers, and the "F" and "P" shapes are similar. When there is partial occlusion or other reasons, F or P may be recognized incorrectly. If the monitoring points corresponding to the adjacent monitoring point identifiers in the merged traveling track sequence have a direct reachable relationship and meet the transit time between the monitoring points, the license plate number error data may be corrected and submitted for manual confirmation or Re-recognize the license plate number after de-noising the image.

In the present embodiment, the correction of the license plate number error data of the vehicle to be inspected ensures the accuracy of the traveling track sequence of the vehicle to be inspected, thereby improving the accuracy of the abnormal behavior recognition of the vehicle to be inspected.

Fourth embodiment:

After acquiring the traveling trajectory sequence of the vehicle to be inspected and the vehicle behavior pattern in the statistical area, it may be determined whether the vehicle to be inspected has an abnormal behavior according to the driving trajectory sequence and the vehicle behavior pattern in the pre-acquired statistical region. Specifically, when determining whether there is an abnormal behavior of the vehicle to be inspected according to the driving trajectory sequence and the vehicle behavior pattern in the pre-acquired statistical region, the trajectory sequence, the reachability relationship between the monitoring points, and the monitoring point may be determined according to the traveling trajectory sequence. The transit time between the vehicles determines whether there is abnormal behavior in the vehicle to be inspected, and can also determine whether the vehicle to be inspected has abnormal behavior according to the traveling trajectory sequence and the driving trajectory template.

The following two explanations are given for the above two ways of judging whether the vehicle to be inspected has abnormal behavior.

First, according to the sequence of the driving track, the reachable relationship between the monitoring points, and the transit time between the monitoring points, when determining whether the vehicle to be inspected has abnormal behavior, the driving track sequence and the monitoring point may be first The reachable relationship and the transit time between the monitoring points, and sequentially determine whether the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle to be inspected have direct reachability relationship and whether the monitoring points are satisfied. Pass time. If the monitoring point corresponding to the adjacent monitoring point identifier does not have a direct reachable relationship, or the monitoring point corresponding to the adjacent monitoring point identifier has a direct reachable relationship but does not satisfy the transit time between the monitoring points, then the determination is made. Wait There is abnormal behavior in the vehicle. Optionally, a confidence interval may be set for the transit time between the monitoring points. When the monitoring point corresponding to the adjacent monitoring point identifier has a direct reachable relationship but the transit time exceeds the confidence interval, the determination is performed. There is abnormal behavior in the vehicle to be tested.

The following describes the principle of judging whether the vehicle to be inspected has abnormal behavior according to the traveling trajectory sequence, the reachability relationship between the monitoring points, and the transit time between the monitoring points.

Under the normal condition of the monitoring point, the vehicle will be collected on the road by multiple monitoring points. The monitoring points corresponding to these monitoring points will form the driving trajectory sequence of the vehicle. Since the space transfer of the vehicle under normal conditions is continuous, that is, the vehicle does not suddenly disappear from a certain place, and suddenly appears in a different place without the vehicle monitoring data along the way, the adjacent monitoring point identification on the traveling track sequence of the vehicle The corresponding monitoring points must have a direct reachable relationship. If there is no direct reachable relationship, the vehicle monitoring data may be abnormal if there is no abnormal monitoring data. In addition, if there is a direct reachable relationship between the monitoring points corresponding to the adjacent monitoring point identifiers on the trajectory sequence of the vehicle, the transit time of the monitoring point corresponding to the vehicle passing the adjacent monitoring point identifier does not satisfy the monitoring point. The transit time between the bits can also determine that the vehicle has abnormal behavior.

According to this method, abnormal behaviors such as vehicle decks, vehicle flops, and occlusion license plate numbers can be identified.

Identify vehicle deck abnormal behavior:

For example, the vehicle monitoring data of the vehicle A to be inspected is sorted, and the traveling trajectory sequence of the vehicle A to be inspected in a certain period of time is obtained as P=(A1, A2, ..., An), and the adjacent monitoring in the traveling trajectory sequence P is sequentially determined. Whether there is a direct reachability relationship between the monitoring points corresponding to the point identifiers and whether the transit time between the monitoring points is met. If P(i,i+1)=0, that is, there is no direct reachability relationship between the monitoring points corresponding to the monitoring point identifiers Ai and Ai+1, it can be recorded as an abnormality; if the monitoring point identifier Ai The monitoring point corresponding to A i+1 has a direct reachable relationship, but the time t(i, i+1) of the monitoring point corresponding to the monitoring point identifiers Ai and A i+1 does not satisfy the monitoring point. The transit time between bits, such as the time t(i, i+1) outside the 95% confidence interval of the transit time between the monitoring points, can also be recorded as an exception. Finally, according to the proportion of abnormal data, you can get the order of all vehicles being decked.

Assuming that the traveling trajectory sequence of the deck vehicle is Pfake=(A1, A2, ..., An), the traveling trajectory sequence of the real license plate vehicle is Preal=(B1, B2, ..., Bn). If the deck vehicle and the real license plate vehicle appear at the same time, the following sequence Pmix=(A1, A2, B1, A3, B2, B3, ..., Bn, An) may be obtained after the vehicle monitoring data is sorted by time. In this case, if the driving route of the deck vehicle and the real license plate vehicle is close, and the time when the vehicle passes the monitoring point corresponding to the adjacent monitoring point identifier meets the transit time between the monitoring points, then the monitoring point is passed. The transit time cannot be found in the vehicle deck, but if it is based on the reachability relationship between the monitoring points, it can be found that the adjacent monitoring point identifiers A2 and B1, B1 and A3, A3 and B2 correspond to the monitoring points. There is no direct reachability relationship, so that the vehicle corresponding to the sequence Pmix has abnormal behavior, and since the sequence Pmix corresponds to the same license plate number, it is concluded that the vehicle corresponding to the Pmix may have a deck abnormal behavior. In addition, if the deck vehicle and the real license plate vehicle do not appear at the same time, the trajectory sequence of the deck vehicle and the real license plate vehicle is combined to be Pmix=(A1, A2, ..., An, B1, B2, ..., Bn), if After the vehicle passes the monitoring point, the time between An and B1 meets the transit time between the monitoring points. At this time, it is impossible to identify whether the vehicle has a deck abnormal behavior by monitoring the transit time between the points, but because the deck vehicle Different from the real license plate vehicle activity range, if the monitoring point corresponding to the monitoring point identifiers An and B1 does not have a direct reachable relationship at this time, it can still be judged that the vehicle corresponding to the Pmix may have a deck abnormal behavior.

Identify vehicle flops and obstruction of license plate number anomalies:

The vehicle monitoring data of the vehicle A to be inspected is sorted, and the traveling trajectory sequence of the vehicle A to be inspected in a certain period of time is obtained as P=(A1, A2, ..., An), and the adjacent monitoring points in the traveling trajectory sequence P are sequentially determined. Identifies whether there is a direct reachability relationship between the corresponding monitoring points and whether the transit time between the monitoring points is met. If it is found that there is no direct reachable relationship between the monitoring points corresponding to the monitoring point identifiers A1 and A2, and the passing time of the vehicle A to be inspected at the monitoring point corresponding to the monitoring point identifiers A1 and A2 is t1 respectively. And t2, at this time, it is possible to take the license plate numbers L=(L1, L2, ..., Ln) of the monitoring points corresponding to the monitored point identifiers A1 and A2 within the time period from t1 to t2, and separately for each license plate. The vehicle monitoring data of the number between the time before t1 and the time after t2 is analyzed. If the vehicle Li is in the trajectory sequence between the time before t1 and the time after t2, if the monitoring point corresponding to the adjacent monitoring point identifier does not have a direct reachable relationship, the vehicle Li is listed as suspected. Suspected vehicle. Of course, if the license plate number of the vehicle Li is not recognized, the vehicle Li is also listed as a suspect vehicle. Finally, image identification is further performed on all vehicle monitoring data of the suspect vehicle. If there are external features such as vehicle type, color, etc., which match the external features of the vehicle A to be inspected, it is considered that the vehicle A to be inspected has a flop or an occlusion number plate, etc. Abnormal behavior, at this time, the vehicle monitoring data of the vehicle A to be inspected can be extracted for manual review.

Secondly, when determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the driving trajectory template, the traveling trajectory sequence of the vehicle to be inspected may be split into a plurality of driving trajectory sub-sequences according to a preset time interval; Then, according to the first monitoring point identifier and the last monitoring point identifier of the driving track subsequence, the driving track class template having the first monitoring point identifier and the last monitoring point identifier is obtained; finally, the driving track subsequence and the obtained driving are calculated. The similarity between the trajectory class templates, if the similarity is less than the second preset value, determining that the vehicle to be tested has an abnormal behavior.

The following is an explanation of whether the vehicle to be inspected has abnormal behavior according to the traveling trajectory sequence and the traveling trajectory class template.

Suppose that there are a total of m driving trajectory templates in a statistical area, namely S={S1, . . . , Sm}, and the vehicle monitoring data of the vehicle A to be inspected is sorted, and the traveling trajectory sequence of the vehicle A to be inspected in a certain time period is obtained. P = (A1, A2, ..., An). At this time, according to a preset time interval, for example, 1 hour, the traveling track sequence P=(A1, A2, . . . , An) of the vehicle A to be inspected is split into a plurality of driving track sub-sequences, such as P1=(A1, A2). ,...,Ai),P2=(Ai+1,Ai+2,...,Aj),......,Pn=(Aj+1,Aj+2,...,An); then according to the split driving track respectively The first monitoring point identifier and the last monitoring point identifier of the sub-sequence Pi are searched for the driving trajectory class template Si having the same first monitoring point identifier and the last monitoring point identifier; finally, the driving trajectory sub-sequence Pi and the driving trajectory are respectively calculated. The similarity between the class templates Si, if the similarity is less than the second preset value, can determine that the vehicle A to be tested has an abnormal behavior. For example, using a typical Hausdorff distance method and the like, the smaller the similarity between the travel track sub-sequence Pi and the travel track type template Si, the greater the possibility that the vehicle to be tested has abnormal behavior.

In this embodiment, according to the traveling trajectory sequence and the vehicle behavior pattern in the pre-acquired statistical region, it is determined whether there is abnormal behavior of the vehicle to be inspected, and the correlation analysis of the vehicle monitoring data collected by the plurality of monitoring points cannot be performed in the prior art. Defects, while solving the problem that can't be long The problem of automatic judgment of abnormal behavior of vehicles in the room increases the recognition accuracy and recognition efficiency of the abnormal behavior of the vehicle to be tested.

Fifth embodiment:

FIG. 6 is a structural block diagram of an apparatus for identifying an abnormal behavior of a vehicle according to a fifth embodiment of the present invention, the identification apparatus comprising:

The first obtaining module 401 is configured to acquire vehicle monitoring data collected by a plurality of monitoring points set in the statistical area, and extract behavior characteristic data of the vehicle to be inspected from the vehicle monitoring data;

The second obtaining module 402 is configured to acquire a traveling trajectory sequence of the vehicle to be inspected according to the behavior characteristic data;

The judging module 403 is configured to determine whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the vehicle behavior pattern in the pre-acquired statistical region.

Optionally, the behavior characteristic data includes a license plate number and a monitoring point identifier, and the second acquiring module is further configured to: according to the license plate number of the vehicle to be inspected and the time sequence of the plurality of monitoring points in the statistical area of the vehicle to be inspected, The monitoring point identification corresponding to the plurality of monitoring points that the vehicle passes in sequence is recorded as the traveling trajectory sequence of the vehicle to be inspected.

Optionally, the identifying device further includes a third acquiring module, configured to acquire a vehicle behavior pattern according to vehicle monitoring data of all vehicles in the pre-acquired statistical area, wherein the vehicle behavior mode comprises: a reachability relationship between the monitoring points , monitoring the transit time between the points and the vehicle's driving trajectory class template.

Optionally, the third obtaining module is further configured to: obtain a traveling trajectory sequence of all the vehicles according to vehicle monitoring data of all vehicles in the statistical area; perform statistical analysis on the traveling trajectory sequence of all the vehicles, obtain a normal driving trajectory sequence, and determine The adjacent monitoring point identifier in the normal driving track sequence; according to the adjacent monitoring point identifier, the monitoring point with direct reachability relationship is obtained, wherein the previous monitoring point indicated by the adjacent monitoring point identifier can be directly reached The next monitoring point in the adjacent monitoring point identifier.

Optionally, the third obtaining module is further configured to calculate, according to the vehicle monitoring data, an average transit time between monitoring points having a direct reachability relationship within a preset time period, and record the average transit time as the pre-predetermined time Set the transit time between the monitoring points in the time period.

Optionally, the identifying device further includes a correction module configured to correct the license plate number error data of the vehicle to be inspected, and further configured to determine, according to the reachable relationship between the monitoring points, the adjacent one of the traveling track sequences of the to-be-tested vehicle Whether there is a direct reachability relationship between the monitoring points corresponding to the monitoring point identifiers; if not, all paths between the monitoring points in the statistical area that do not have direct reachability relationship are obtained, wherein the paths are sequentially passed A sequence representation of the monitoring point identifiers corresponding to the monitoring points; obtaining monitoring point identifiers for monitoring abnormal monitoring data in all paths according to the monitoring point identifiers in all the paths, wherein the abnormal monitoring data includes at least: the monitored The license plate number does not match the vehicle registration information, the license plate number registration information, the monitoring point corresponding to the adjacent monitoring point identifier in the path does not have a direct reachability relationship, and the monitoring point with direct reachability in the path does not satisfy the monitoring point. The transit time between the two; the monitoring point identifier that monitors the abnormal monitoring data is merged into the line of the vehicle to be inspected In the trajectory sequence, it is determined whether the monitoring point corresponding to the adjacent monitoring point identifier in the merged traveling trajectory sequence satisfies the transit time between the monitoring points; if satisfied, and the monitored license plate number and the license plate of the vehicle to be inspected If the similarity of the number plate is greater than the first preset value, the monitored license plate number is corrected.

Optionally, the judging module is further configured to: determine, according to the trajectory sequence, the reachability relationship between the monitoring points, and the transit time between the monitoring points, whether the vehicle to be tested has abnormal behavior; and/or according to the trajectory sequence And the driving track class template to determine whether the vehicle to be inspected has abnormal behavior.

Optionally, the judging module includes a first judging unit configured to sequentially determine the neighboring positions of the to-be-tested vehicle trajectory according to the traveling trajectory sequence, the reachability relationship between the monitoring points, and the transit time between the monitoring points. Whether the monitoring point corresponding to the monitoring point identifier has a direct reachability relationship and whether the transit time between the monitoring points is met; if the monitoring point corresponding to the adjacent monitoring point identifier does not have a direct reachable relationship, or adjacent monitoring If the monitoring point corresponding to the point identifier has a direct reachability relationship but does not satisfy the transit time between the monitoring points, it is determined that the vehicle to be tested has an abnormal behavior.

Optionally, the determining module further includes a second determining unit configured to split the traveling trajectory sequence of the to-be-tested vehicle into a plurality of driving trajectory sub-sequences according to a preset time interval; and the first monitoring position according to the driving trajectory sub-sequence Identification and last monitoring point identification, obtaining the first monitoring point mark Knowing the driving trajectory class template identified by the last monitoring point; calculating the similarity between the driving trajectory subsequence and the acquired driving trajectory class template, and if the similarity is less than the second preset value, determining that the vehicle to be inspected has an abnormal behavior .

Embodiments of the present invention also provide a storage medium. Optionally, in the embodiment, the foregoing storage medium may be configured to store program code for performing the following steps:

Step S1: acquiring vehicle monitoring data collected by a plurality of monitoring points set in the statistical area, and extracting behavior characteristic data of the vehicle to be inspected from the vehicle monitoring data;

Step S2, acquiring a travel trajectory sequence of the vehicle to be inspected according to the behavior characteristic data;

In step S3, it is determined whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the vehicle behavior pattern in the statistical area acquired in advance.

Optionally, in this embodiment, the foregoing storage medium may include, but not limited to, a USB flash drive, a Read-Only Memory (ROM), a Random Access Memory (RAM), a mobile hard disk, and a magnetic memory. A variety of media that can store program code, such as a disc or a disc.

The above is a preferred embodiment of the present invention, and it should be noted that those skilled in the art can also make several improvements and refinements without departing from the principles of the present invention. These improvements and refinements are also within the scope of the present invention. Inside.

Industrial applicability

The embodiment of the invention is applied to the field of intelligent traffic monitoring, and solves the defect that the vehicle monitoring data collected by the plurality of monitoring points cannot be correlated in the prior art by analyzing the traveling trajectory sequence of the vehicle to be tested, and solves the problem at the same time. The problem of not being able to automatically judge the abnormal behavior of the vehicle over a long period of time increases the recognition accuracy and recognition efficiency of the abnormal behavior of the vehicle.

Claims

A method for identifying abnormal behavior of a vehicle, the method comprising:

Obtaining vehicle monitoring data collected by a plurality of monitoring points set in the statistical area, and extracting behavior characteristic data of the vehicle to be inspected from the vehicle monitoring data;

Obtaining a travel trajectory sequence of the vehicle to be inspected according to the behavior characteristic data;

Determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the vehicle behavior pattern in the statistical region acquired in advance.
The identification method according to claim 1, wherein the behavior characteristic data includes a license plate number and a monitoring point identification, and the acquiring the traveling trajectory sequence of the vehicle to be inspected according to the behavior characteristic data comprises:

Recording, according to the license plate number of the vehicle to be inspected and the time sequence of the plurality of monitoring points in the statistical area of the vehicle to be inspected, recording the monitoring point identifier corresponding to the plurality of monitoring points that the vehicle to be inspected sequentially passes as the waiting Check the vehicle's travel trajectory sequence.
The identification method according to claim 2, wherein the determining whether the vehicle to be inspected has an abnormal behavior according to a traveling trajectory sequence of the vehicle to be inspected and a vehicle behavior pattern in the statistical region acquired in advance The identification method further includes:

Obtaining a vehicle behavior pattern according to pre-acquired vehicle monitoring data of all vehicles in the statistical area, wherein the vehicle behavior pattern comprises: a reachability relationship between monitoring points, a transit time between monitoring points, and a vehicle Driving track class template.
The identification method according to claim 3, wherein the reachability relationship between the monitoring points is determined as follows:

Acquiring a sequence of travel trajectories of all the vehicles according to vehicle monitoring data of all vehicles in the statistical area;

Performing statistical analysis on the traveling trajectory sequences of all the vehicles to obtain a normal driving trajectory sequence, and determining adjacent monitoring point identifiers in the normal driving trajectory sequence;

Obtaining a monitoring point with a direct reachability relationship according to the neighboring monitoring point identifier, wherein the previous monitoring point indicated by the adjacent monitoring point identifier can directly reach the adjacent monitoring point identifier A monitoring point.
The identification method according to claim 4, wherein the transit time between the monitoring points is determined as follows:

Calculating, according to the vehicle monitoring data, an average transit time between monitoring points having a direct reachable relationship in a preset time period, and recording the average transit time as a monitoring point in the preset time period. The transit time between.
The identification method according to claim 4, wherein after the obtaining the travel trajectory sequence of the vehicle to be inspected according to the behavior characteristic data, the identification method further comprises:

Correcting the license plate number error data of the vehicle to be inspected, specifically including:

Determining, according to the reachability relationship between the monitoring points, whether there is a direct reachable relationship between the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle to be inspected;

If not, all the paths between the monitoring points in the statistical area that do not have a direct reachability relationship are obtained, where the path is represented by a sequence consisting of monitoring point identifiers corresponding to the monitoring points in sequence;

And acquiring, according to the monitoring point identifiers in the all the paths, the monitoring point identifiers of the abnormal monitoring data that are monitored in the all the paths, where the abnormal monitoring data includes: the monitored license plate number does not match the vehicle registration information, No license plate number registration information, monitoring points corresponding to adjacent monitoring point identifiers in the path do not have direct reachability relationship and path The monitoring point with direct reachability does not satisfy the transit time between monitoring points;

Merging the monitoring point identifier of the monitored abnormality monitoring data into the traveling trajectory sequence of the vehicle to be inspected, and determining whether the monitoring point corresponding to the adjacent monitoring point identifier in the merged traveling trajectory sequence satisfies the monitoring point Between passages;

If it is satisfied, and the similarity between the monitored license plate number and the license plate number of the vehicle to be inspected is greater than the first preset value, the monitored license plate number is corrected.
The identification method according to claim 4, wherein the determining whether the vehicle to be inspected has an abnormal behavior according to a traveling trajectory sequence of the vehicle to be inspected and a vehicle behavior pattern in the statistical region acquired in advance includes :

Determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence, the reachability relationship between the monitoring points, and the transit time between the monitoring points; and/or

Determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the driving trajectory class template.
The identification method according to claim 7, wherein said determining whether said vehicle to be inspected has an abnormal behavior according to said traveling trajectory sequence, a reachable relationship between monitoring points, and a transit time between monitoring points ,include:

Determining, according to the traveling trajectory sequence, the reachable relationship between the monitoring points, and the transit time between the monitoring points, whether the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling trajectory sequence of the vehicle to be inspected are sequentially determined whether Have a direct reachability relationship and whether the transit time between monitoring points is met;

If the monitoring point corresponding to the adjacent monitoring point identifier does not have a direct reachable relationship, or the monitoring point corresponding to the adjacent monitoring point identifier has a direct reachable relationship but does not satisfy the transit time between the monitoring points, then the determination is made. The vehicle to be tested has an abnormal behavior.
The identification method according to claim 7, wherein the determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the driving trajectory class template comprises:

Disassembling the traveling trajectory sequence of the to-be-tested vehicle into a plurality of driving trajectory sub-sequences according to a preset time interval;

Obtaining a driving track class template having the first monitoring point identifier and the last monitoring point identifier according to the first monitoring point identifier and the last monitoring point identifier of the driving track subsequence;

Calculating a similarity between the travel track subsequence and the acquired travel track class template, and if the similarity is less than a second preset value, determining that the vehicle to be tested has an abnormal behavior.
An identification device for abnormal behavior of a vehicle, the identification device comprising:

a first acquiring module, configured to acquire vehicle monitoring data collected by a plurality of monitoring points set in the statistical area, and extract behavior characteristic data of the to-be-tested vehicle from the vehicle monitoring data;

a second acquiring module, configured to acquire a traveling trajectory sequence of the to-be-tested vehicle according to the behavior characteristic data;

The determining module is configured to determine whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the vehicle behavior pattern in the statistical region acquired in advance.
The identification device according to claim 10, wherein the behavior characteristic data includes a license plate number and a monitoring point identification, and the second acquisition module is further configured to: according to the license plate number of the vehicle to be inspected and the to-be-tested The time sequence of the plurality of monitoring points in the statistical area is recorded by the vehicle, and the monitoring point identifier corresponding to the plurality of monitoring points passing through the vehicle to be inspected is recorded as the traveling track sequence of the vehicle to be inspected.
The identification device according to claim 11, wherein the identification device further comprises a third acquisition module configured to acquire a vehicle behavior pattern according to pre-acquired vehicle monitoring data of all vehicles in the statistical area, wherein Vehicle behavior patterns include: the reachability relationship between monitoring points, the transit time between monitoring points, and the trajectory template of the vehicle.
The identification device according to claim 12, wherein the third acquisition module is further configured to acquire a travel trajectory sequence of all the vehicles according to vehicle monitoring data of all vehicles in the statistical area; The trajectory sequence is statistically analyzed to obtain a normal trajectory sequence, and the adjacent monitoring point identifiers in the normal driving trajectory sequence are determined; according to the adjacent monitoring point identifiers, the monitoring points with direct reachability relationship are obtained. The previous monitoring point indicated by the adjacent monitoring point identifier may directly reach the next monitoring point in the adjacent monitoring point identifier.
The identification device according to claim 13, wherein the third obtaining module is further configured to calculate, according to the vehicle monitoring data, an average traffic between monitoring points having a direct reachability relationship within a preset time period Time, and record the average transit time as the transit time between the monitoring points within the preset time period.
The identification device according to claim 13, wherein the identification device further comprises a correction module configured to correct the license plate number error data of the vehicle to be inspected, further configured to be based on the monitoring point a reachable relationship, determining whether there is a direct reachability relationship between the monitoring points corresponding to the adjacent monitoring point identifiers in the traveling track sequence of the vehicle to be inspected; if not, obtaining the statistical area does not have a direct All the paths between the monitoring points of the relationship, wherein the path is represented by a sequence consisting of monitoring point identifiers corresponding to the monitoring points in sequence; and all the paths are obtained according to the monitoring point identifiers in all the paths The monitoring point identifier of the abnormal monitoring data is monitored, wherein the abnormal monitoring data includes at least: the monitored license plate number does not match the vehicle registration information, the license plate number registration information, and the monitoring corresponding to the adjacent monitoring point identifier in the path The point does not have a direct reachable relationship and the monitoring point with direct reachability in the path is not satisfied. The transit time between the monitoring points; the monitoring point identifier of the monitored abnormal monitoring data is merged into the traveling track sequence of the vehicle to be inspected, and it is determined that the adjacent monitoring point identifiers in the merged traveling track sequence correspond to Whether the monitoring point meets the transit time between the monitoring points; if it is satisfied, and the similarity between the monitored license plate number and the license plate number of the vehicle to be inspected is greater than the first preset value, the monitored license plate will be The number is corrected.
The identification device according to claim 13, wherein the determining module is further configured to determine the waiting according to the traveling trajectory sequence, the reachable relationship between the monitoring points, and the transit time between the monitoring points Detecting whether there is abnormal behavior of the vehicle; and/or determining whether the vehicle to be inspected has an abnormal behavior according to the traveling trajectory sequence and the driving trajectory type template.
The identification device according to claim 16, wherein the determination module comprises a first determination unit configured to set a travel time between the sequence of the travel trajectory, the monitoring point, and the travel time between the monitoring points, Determining, in sequence, whether the monitoring point corresponding to the adjacent monitoring point identifier in the traveling track sequence of the vehicle to be inspected has a direct reachability relationship and whether the transit time between the monitoring points is met; if the adjacent monitoring point identifier corresponds to If the monitoring point does not have a direct reachability relationship, or the monitoring point corresponding to the adjacent monitoring point identifier has a direct reachable relationship but does not satisfy the transit time between the monitoring points, it is determined that the vehicle to be tested has an abnormal behavior.
The identification device according to claim 16, wherein the determining module further comprises a second determining unit configured to split the traveling trajectory sequence of the vehicle to be inspected into a plurality of driving trajectories according to a preset time interval And obtaining a driving track class template having the first monitoring point identifier and the last monitoring point identifier according to the first monitoring point identifier and the last monitoring point identifier of the driving track subsequence; and calculating the driving track And a similarity between the subsequence and the acquired driving trajectory class template, and if the similarity is less than the second preset value, determining that the vehicle to be inspected has an abnormal behavior.